摘要 |
PROBLEM TO BE SOLVED: To provide a high-quality speech encoding method and system which lessens the occurrence of the local abnormal sound of decoding sounds. SOLUTION: This system has a driving vector forming means 13 which forms plural driving vectors, a first distortion calculating section 23 which calculates the distortion relating to the waveforms defined between the signal to be encoded which is determined from input speech by each of the respective driving vectors and the synthesized vector determined from the driving vectors as first distortion, a second first distortion calculating section 24 which calculates the second distortion different from the first distortion relating to the waveforms defined between the signal to be encoded by each of the respective driving vectors and the synthesized vector determined from the driving vectors, an evaluation value calculating section 29 which calculates the prescribed evaluation value for search by using the first distortion and the second distortion by each of the respective driving vectors and a search means 20 which selects the driving vectors to minimize the evaluation value for search and outputs the codes previously made correspondent to the selected driving vectors. |